Google DeepMind CEO pushes for AI watchdog in Washington

Google DeepMind CEO Demis Hassabis is lobbying U.S. lawmakers for a permanent, FINRA-style AI watchdog to test frontier models for national-security risks before release, potentially making approval a prerequisite for U.S. market entry. The discussion comes as the head of the Trump administration’s AI security agency reportedly resigns, highlighting regulatory uncertainty in Washington. For Alphabet, the broader market angle is mixed: tighter regulation could slow rivals and formalize oversight, but it also raises compliance risk for AI product rollout. Separately, reports that Google may be developing a specialized AI chip to improve Gemini efficiency helped lift the stock. The article also notes competitive pressure from Chinese models and internal execution bottlenecks at DeepMind, underscoring the race to bring AI products to market.